1 of 35

SpaDES�An open platform for transparent, science-informed policy, decision making, and co-production

Eliot McIntire, Research Scientist�CFS – Pacific�Alex M. Chubaty,Steve Cumming, Ceres Barros, Tatiane Micheletti, Ian Eddy, Celine Boisvenue, Frances E. C. Stewart, and many others

2 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

3 of 35

SpaDES: what and how

A platform for data and models built in R, the data language
A conductor of data and models
A standard so that the conductor has something to work with
Analogous to “an Android” or “an iPhone”

Apps or “modules” are the science
Simple (“a clock app”) or rich (“google maps”)

We can make collections of “modules” for any purpose

Cumulative effects, integrated management, forest landscape planning, decision support, risk assessments, resilience, co-production …

SELES

Ecological modeler

Teacher

75 students

Some of the abstractions in SELES were fantastic, but just not broad enough to handle all needs

Web development, parameter estimation, GIS, modern statistics, data science

Least cost path surface 🡪 Andrew Fall could make this, I tried – 2 weeks, almost there … kept finding edge cases … finally looked at R – there was a package, somebody else developed, I installed it and in 5 minutes had my least cost path. I interrupted my SELES model every year, send out GIS layers to R, sent them back in; this worked. I made an R package called “SELES” so that this sort of thing would be easier next time. We published the paper Caribou, forestry, fires in Quebec. It was painful and slow. I started looking around for what else existed … many platforms … I hired a research associate to evaluate these … Essentially none of them had the power and modern capacity, without the need to become a programmer in a foreign language

4 of 35

What other platforms?

5 of 35

Web page

spades.predictiveecology.org

6 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

7 of 35

Nimble Management

Modelling tool: set of R Packages
Philosophy: PERFICT

Predictive

Ecology based on

Reusable,

Freely available,

Interoperable models built around a

Continuous Workflow that is

Tested automatically

SpaDES

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

8 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

This circle can’t handle the real world complexity. We don’t want to wait until the cycle is complete

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

9 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the problem changes, it might directly affect our assessment, or forecasts, or decisions

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

10 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the data change, it might directly affect our analyses or the problem

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

11 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the models change, it might directly change the data we need and decisions

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

12 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the analyses change …

13 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the forecasts change …

14 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the assessments change …

15 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

If the decisions change …

16 of 35

Nimble management

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

From any change, we want to be able to make updates where they are needed

17 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

18 of 35

Cumulative effects of natural and anthropogenic disturbances under climate change in Western Canada:�� Assessing the synergies & trade-offs among ��long-term conservation of priority species and places, carbon sequestration, and land management��

19 of 35

Drivers 🡪 � Indicators 🡪 � Targets/Thresholds/Limits 🡪� Risks

Real world problems:

Many Drivers & Changing
Many Indicators & Changing
Costing of crossing targets/thresholds/limits

Cumulative Effects & Risk

Climate change

20 of 35

Drivers 🡨 � Indicators 🡨� Targets/Thresholds/Limits 🡨� Risks�

We don’t need only ask, what are effects of drivers on indicators 🡪 we can affect the drivers (harvesting, pests, fires, O&G etc.) knowing the indicators/limits 🡪 through scenarios/optimization

Cumulative Effects & Risk

Climate change

21 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

22 of 35

Scenarios – �identify trade-offs/synergies

Partitioning “effects” on indicators (e.g., caribou, biodiversity, carbon, jobs) among causes of change (climate, timber, roads, fire, fire management, other) in light of Indigenous management and co-production
Deriving trade-offs & synergies between favouring Timber, Caribou, SAR, Carbon, Fire Management, (Insect management) 🡪 inform these conflicting policy drivers
Building a re-usable foundation for other groups/projects
Risks of failure/crossing limits

23 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

24 of 35

Western Boreal

At least 14 different lead developers on these

25 of 35

BC Regional Integrated Assessment

Reuse the parts we can, simplify as needed, build a few new pieces

26 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

27 of 35

Landscape Planning

Simple when needed
Complex sometimes
Some issues are not present in some places
Some places have unique issues�
Solutions must be nimble, flexible
Tools must be “close” to data yet “available” to people/ communities

28 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

29 of 35

Co-production requires nimbleness

New data/�science

New analyses/�models

New algorithms/�models

New indicators/�factors

New politics/�economics

New concerns

New forecasts

30 of 35

Co-production – Web apps

For stakeholder involvement, and decisions, assumption viewing/changing etc. allowing for transparent process from models and data�https://dev01.01101.io/rstudio/ (private)
http://LandWeb.ca a proof of concept, it is still slower than desired
It takes hours to weeks to make web apps, as we can re-use many components

31 of 35

Co-production – modules

We are building a proof of concept IEK Caribou Habitat Suitability model based on interviews (Polfus et al 2014 ) NW BC - with Taku River Tlingit First Nation
Modules are simple to make
The content is the hard part

32 of 35

Outline

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience

33 of 35

Ecological Resilience

Build scenarios
Need modules that can give information about indicators
Define resilience

E.g., compare “potential productivity” calculated in a BGC module with the “Climate Sensitive G&Y” module to determine whether we have created “productivity resilience”
E.g., Caribou population size can withstand forecasted disturbances (avg. λ > 1.025)

34 of 35

SpaDES – Why this works

Focussed around R and data
Nimble by design
Many of us already know the language ( R ) for the data analysis
Brings many values:

Transparency, reproducibility, re-usability, continuous workflow

35 of 35

Thank you

SpaDES – what/how
Nimble management
Cumulative effects
Scenarios
Cross-disciplines
Landscape planning
Co-production
Ecological resilience